-
Notifications
You must be signed in to change notification settings - Fork 0
/
Bank_Marketing.py
330 lines (240 loc) · 13 KB
/
Bank_Marketing.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
# coding: utf-8
# # Binary Classification Model
# The goal of this code is to train and compare the performance of three popular classifiers (e.g., Logistic Regression, Random Forests, and Boosting) on a publicly available data set.
# Data : The data is related with direct marketing campaigns of a Portuguese banking institution. The marketing campaigns were based on phone calls. Often, more than one contact to the same client was required, in order to access if the product (bank term deposit) would be ('yes') or not ('no') subscribed.
# Goal: Predict " has the client subscribed a term deposit?"
# In[1]:
from sklearn.model_selection import cross_val_score
import numpy as np
from sklearn.ensemble import RandomForestClassifier
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder
from sklearn.metrics import roc_curve
from sklearn.metrics import auc
import matplotlib.pylab as plt
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import average_precision_score
from sklearn.metrics import precision_recall_curve
from sklearn.ensemble import GradientBoostingClassifier
import multiprocessing as mp
from sklearn.feature_selection import GenericUnivariateSelect
from sklearn.feature_selection import f_classif
from sklearn.ensemble import ExtraTreesClassifier
from sklearn import metrics
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix
import itertools
from sklearn.preprocessing import OneHotEncoder
get_ipython().magic(u'matplotlib inline')
# In[2]:
def roc_auc_rfc(clf,X_test,Y_test):
target_predict = clf.predict_proba(X_test)
fpr_rfc, tpr_rfc, _ = roc_curve(Y_test, target_predict[:,1])
return auc(fpr_rfc, tpr_rfc)
def accuracy(clf,X_test,Y_test):
target_predict = clf.predict(X_test)
return metrics.accuracy_score(Y_test, target_predict)
def roc_curve_plot(clfs,clf_names,X_test,Y_test):
for clf,name in zip(clfs,clf_names):
target = clf.predict_proba(X_test)
fpr, tpr, _ = roc_curve(Y_test, target[:,1])
roc_auc = auc(fpr, tpr)
# plot the roc-auc
plt.plot(fpr, tpr, label=name +' ROC curve (area = %0.3f)' % roc_auc)
plt.plot([0, 1], [0, 1], 'k--') # random predictions curve
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.0])
plt.xlabel('False Positive Rate or (1 - Specifity)')
plt.ylabel('True Positive Rate or (Sensitivity)')
plt.title('Receiver roc_aucOperating Characteristic')
plt.legend(loc="lower right")
plt.show()
def precious_recall_plot(clfs,clf_names,X_test,Y_test):
for clf,name in zip(clfs,clf_names):
target = clf.predict_proba(X_test)
precision, recall, _ = precision_recall_curve(Y_test, target[:,1])
prc_auc = average_precision_score(Y_test, target[:,1])
plt.plot(recall, precision, label=name+' (area = %0.3f)' % prc_auc)
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.0])
plt.xlabel('Recall')
plt.ylabel('Precision')
plt.title('Precision-Recall curve')
plt.legend(loc="lower right")
plt.show()
def plot_confusion_matrix(cm, classes,
normalize=False,
title='Confusion matrix',
cmap=plt.cm.Blues):
"""
This function prints and plots the confusion matrix.
Normalization can be applied by setting `normalize=True`.
"""
plt.imshow(cm, interpolation='nearest', cmap=cmap)
plt.title(title)
plt.colorbar()
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, rotation=45)
plt.yticks(tick_marks, classes)
if normalize:
cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
print("Normalized confusion matrix")
else:
print('Confusion matrix, without normalization')
print(cm)
thresh = cm.max() / 2.
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, cm[i, j],
horizontalalignment="center",
color="white" if cm[i, j] > thresh else "black")
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
# # Reading and preprocessing data
# We first start with reading and pre-processing the data. We need to convert the categorical variables into numbers. We used Label Encoder, which is condense but put varialbes in order with might be misleading as sequantial variables are closer in distance. An alternative would be panda dummies, which is more sparse.
# Data Source: https://archive.ics.uci.edu/ml/datasets/Bank+Marketing
#
# In[3]:
core_count=core_number= mp.cpu_count()
df =pd.read_csv('bank-full.csv',sep=';')
# encode categorical variables
cat_cols = ['job','marital','education','default','loan','contact','day','month','housing','poutcome','y']
for var in cat_cols:
number = LabelEncoder()
df[var] = number.fit_transform(df[var].astype('str'))
features = list(df.columns)
features.remove('y')
# # Data Exploration and Feature Extraction
# The dimension of feature space is not high, yet we start with exploring the data see what features explain the outcome better than others.
# # Feature variance
# Lets see the variance of each feature variable first
# In[4]:
for col in features:
print col,np.var(df[col])
# # Univariate Statistical test
# We run a one way statistical test for each feature being devided based on the target variable to see if the difference between to divide is statisitcally significant.
# In[5]:
sel=GenericUnivariateSelect(f_classif, mode='percentile', param=1e-05)
sel.fit_transform(df[features], df['y'])
features_significance =pd.DataFrame(index=features)
features_significance['score']=sel.scores_
features_significance['p_value']=sel.pvalues_
features_significance = features_significance.sort_values(['score'],ascending=False)
selected_features=features_significance[features_significance.p_value < 0.05].index
print features_significance
# # Enthropy Based Feature Extraction
# Another way of feature selection is based on tree-based classifiers. Since in tree-based classifers, decision trees are used to divide the data to incrase impurity, the features that are mostly selected for dividing the data should be more informative.
# We use ExtraTreesclassifiers mainly because they are cheaper to train.
# We used all features and did not do a dimension reduction transformation (e.g., PCA) as the size of the data can easily be handled.
# In[14]:
clf_ExtraTrees = ExtraTreesClassifier()
clf_ExtraTrees = clf_ExtraTrees.fit(df[features], df['y'])
importances = clf_ExtraTrees.feature_importances_
importance = pd.DataFrame(importances, index=features,
columns=["Importance"])
importance["Std"] = np.std([tree.feature_importances_
for tree in clf_ExtraTrees.estimators_], axis=0)
x = range(importance.shape[0])
y = importance.ix[:, 0]
yerr = importance.ix[:, 1]
importance= importance.sort_values(['Importance'],ascending=[False])
# Print the feature ranking
print("Feature ranking:")
print importance
# Plot the feature importances of the forest
fig=plt.figure()
plt.title("Feature importances")
plt.bar(range(df[features].shape[1]), importance["Importance"],
color="r", yerr=importance["Std"], align="center")
plt.xticks(range(df[features].shape[1]), importance.index,rotation=90)
plt.xlim([-1, df[features].shape[1]])
plt.show()
# In[7]:
X_train, X_test, Y_train, Y_test = train_test_split(df[features],
df['y'], test_size=0.2, random_state=20)
# # Logistic Regression
# As the baseline, we use Logistic Regression
# In[8]:
clf_Logit = LogisticRegression()
clf_Logit.fit(X_train, Y_train)
print 'roc_auc: ',roc_auc_rfc(clf_Logit,X_test,Y_test)
print 'accuracy score:', accuracy(clf_Logit,X_test,Y_test)
# # Random Forests
# Random forests often outperform LR models. They are quaite robust to missing values, irrelevant features, and outliers. They do not need that much tuning and their paramters are streight forward (e.g., larger number of estimators improve the accuracy, and there is not much difference between different criteria)
# In[9]:
clf_RFC = RandomForestClassifier(n_estimators=1000,
max_features='auto',
n_jobs=core_count)
clf_RFC = clf_RFC.fit(X_train, Y_train)
print 'roc_auc: ',roc_auc_rfc(clf_RFC,X_test,Y_test)
print 'accuracy score:', accuracy(clf_RFC,X_test,Y_test)
# # Gradient Boosting
# Gradient Boosting often slightly outperform RFCs, yet their hyper paramters are more diffult to tune. We used Gridsearch and cross validaiton to tune the paramaters to show how this could be done. This can further be improved though.
# The detail of tuning GBM can be found at: https://www.analyticsvidhya.com/blog/2016/02/complete-guide-parameter-tuning-gradient-boosting-gbm-python/
# In[10]:
param_test1 = {'n_estimators':range(20,200,10)}
gsearch1 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate=0.1, min_samples_split=500,
min_samples_leaf=50,max_depth=8,max_features='sqrt', subsample=0.8,random_state=10),
param_grid = param_test1, scoring='roc_auc',n_jobs=core_count,iid=False, cv=5)
gsearch1.fit(X_train,Y_train)
gsearch1.best_params_, gsearch1.best_score_
# In[11]:
#Grid seach on subsample and max_features
param_test2 = {'max_depth':range(5,20,2), 'min_samples_split':range(100,1001,200)}
gsearch2 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate=0.1, n_estimators=60,
max_features='sqrt', subsample=0.8, random_state=10),
param_grid = param_test2, scoring='roc_auc',n_jobs=4,iid=False, cv=5)
gsearch2.fit(X_train,Y_train)
gsearch2.best_params_, gsearch1.best_score_
# In[12]:
#Grid seach on subsample and max_features
param_test3 = {'min_samples_split':range(500,2100,200), 'min_samples_leaf':range(30,71,10)}
gsearch3 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate=0.1, n_estimators=60,max_depth=9,
max_features='sqrt', subsample=0.8, random_state=10),
param_grid = param_test3, scoring='roc_auc',n_jobs=4,iid=False, cv=5)
gsearch3.fit(X_train,Y_train)
gsearch3.best_params_, gsearch3.best_score_
# In[13]:
param_test4 = {'max_features':range(2,8,2)}
gsearch4 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate=0.1, n_estimators=60,max_depth=9,
min_samples_split=1200, min_samples_leaf=60, subsample=0.8, random_state=10),
param_grid = param_test4, scoring='roc_auc',n_jobs=4,iid=False, cv=5)
gsearch4.fit(X_train,Y_train)
gsearch4.best_params_, gsearch3.best_score_
# In[14]:
param_test5 = {'subsample':[0.6,0.7,0.75,0.8,0.85,0.9]}
gsearch5 = GridSearchCV(estimator = GradientBoostingClassifier(learning_rate=0.1, n_estimators=60,max_depth=9,
min_samples_split=1200, min_samples_leaf=60, subsample=0.8, random_state=10, max_features=7),
param_grid = param_test5, scoring='roc_auc',n_jobs=4,iid=False, cv=5)
gsearch5.fit(X_train,Y_train)
gsearch5.best_params_, gsearch3.best_score_
# In[15]:
clf_GBM= GradientBoostingClassifier(learning_rate=0.005, n_estimators=1500,max_depth=16, min_samples_split=1000,
min_samples_leaf=50, subsample=0.85, random_state=10, max_features=7,
warm_start=True)
clf_GBM = clf_GBM.fit(X_train, Y_train)
print 'roc_auc: ',roc_auc_rfc(clf_RFC,X_test,Y_test)
print 'accuracy score:', accuracy(clf_RFC,X_test,Y_test)
# # Comparing the Performance of Classifiers
# # ROC Curve
# An ideal classifier would stick to (1,0) with auc=1, a random classifer would be the line from (0,0) to (1,1).
# In[16]:
clfs=[clf_Logit,clf_RFC,clf_GBM]
clfs_names=['Logistic Regression','Random Forests','Gradient Boosting']
roc_curve_plot(clfs,clfs_names,X_test,Y_test)
# # Precisiou-Recall Curve
# A high preciuos low recall classier would detect few positive most of which have been classified correctly. A low preciuos high recall classifier would detect a large number of positive many of which have been missclassified.
# Ideally a perfect classifier is high recall high preciouse. However, the trade-off between precious recall depend on the applicaiton. For example, for spam detect we want to have high precious ( we don't want our important emails being mistakenly transferred to spam folder). However, for detecting a HIV we need to have a high recall classifier, so we do not miss any probable positive case.
# In[17]:
precious_recall_plot(clfs,clfs_names,X_test,Y_test)
# # Confusion Matrix
# In[18]:
Y_pred=clf_GBM.predict(X_test)
cnf_matrix = confusion_matrix(Y_test, Y_pred)
np.set_printoptions(precision=2)
# Plot non-normalized confusion matrix
plt.figure()
plot_confusion_matrix(cnf_matrix, classes=['Subscribed','Not subscribed'],
title='Confusion matrix, without normalization')
# In[ ]: